In part one I analyzed the electricity requirements, so now let's start putting this in terms of components, and what it all costs. The components that make up the entire system are: the solar panels, the battery, solar charge controller and the inverter (to make 230V AC). My goal for this system was to do it as cost effective as possible. Note: I said cost effective, not cheapest. I wanted to make sure that the equipment could handle the needs, and could deal with the power I wanted without under-sizing or cutting costs when it comes to safety or reliability. What I didn't want to do was just to run out and buy Victron or Schneider equipment, maybe with SimpliPhi batteries and call it a day. With that said, I also realize that the options I went for are not necessarily feasible for everyone. In the mean time other some options have become available that are more readily accessible for everyone (especially when it comes to batteries) so I will mention those as well. This blog entry gets a bit technical at times, but this is pretty much unavoidable with DIY set-ups.
As you may remember from the previous post, I used this tool to size the solar array and battery pack. The focus was on getting at least a few days of autonomy and a solar system that was sized appropriately to be able to make it from (and including) February to the beginning of November. This turned out to be feasible with 10kW of solar panels and around 28kWh of battery. I also knew I could easily expand the battery later on if needed it, perhaps I will do that once I buy an EV.
 |
| Sunny winter's day. Not really possible to generate a lot of power... |
Solar panels
Let's say that when I bought them, solar panels weren't very common in Finland. They are still much less in use here compared to the rest of Europe. I understand why: when you make the most power in summer, you need it the least. In winter, when you would need the power, you can't generate it. This means that residential solar installations like you have in central and southern Europe are practically non existent. This also means that buying solar panels and other equipment locally is a) expensive and b) has limited options or is not available at all.
I opted to build the 10kW array using twenty 500W monocrystalline panels. These panels are big; 96 cell measuring 1.3 meter wide by almost 2 meters tall and are heavy - 26.5kg each. They are usually aimed at more commercial installations, but I liked the idea of having less panels overall (simplifies wiring) and I like their aesthetics. Based on some initial calculations, I would also be able to easily make a ground mount for these panels with readily available material and minimal waste.
Of course, getting these anywhere in Europe was, and still is, almost impossible. If you do find them, they're expensive. Yes, I could have given up here and just gone with 380W panels or something... but I decided to import them directly from China - this was my source. Total cost including taxes and shipping to my door was 4141€. This comes down to €0.41/Watt, which is even today still cheaper than retail price on commonly available panels here in Finland. Alternatively, there are a few suppliers on Alibaba that have stock in Europe now, for example this one that could be a good option.
Now would I do that again today? Personally, I would - but I probably won't recommend going this route to others today. Shipping costs have increased substantially after the pandemic, and waiting for three months for the panels to arrive by boat is not for everyone. If you're in the States, you have to add tariffs on top as well (Europe doesn't have those) meaning any potential cost savings are gone. Used panels are often a very good option in countries like the States or Australia, and definitely something to look into! Having said all that, I'm very happy with the panels. Not only do they perform as they are supposed to, they look good: the 'massiveness' of them together with how it simplified (in my case) the ground mount design is a great bonus. Having a large panel also simplifies wiring, needs less mounting hardware, etc. They went through several storms in the mean time, and a very long and cold winter without any issues.
I designed my own ground mount system for these panels, holding them at an angle of about 65 degrees (which as you might remember was optimal for autumn/spring). The system is made using pressure treated wood and concrete pillars. It's not easy to dig deep here where I am due to bedrock, so screw poles or other commonly used anchoring methods are difficult and expensive. The pillars have a 'foot', so they can be placed below ground and gravel placed on top adds additional balance. They're 60kg each, and six of them form the base on which the wooden structure is built.
The panels are held in place with a selection of different L-brackets screwed into the wood, so this eliminated the need for dedicated solar panel mounting rails, which aren't exactly cheap here either. Wood on the other hand is readily available and pretty cheap, so no problems with the rest of the ground mount. The wood has pretty much perfect dimensions for these panels, with no cut-offs or other waste. The total cost for each mount (capable of holding 6 panels):
- Timber was 50x100x4200mm, in total 10 pieces used at a cost of €11 a piece delivered
- The 6 concrete posts (60kg each) and accompanying M20 pillar shoes, for a cost of €30 each delivered
- Fasteners and L brackets, maybe for a total of €30
Total: €320 including VAT and delivery. Time to build: a few hours; digging in and leveling the pillars not included. It helps to time the build of the mount in autumn, since a lot of this wood and pillars is typically used to build decks, veranda's etc. Since no one builds these things in Autumn, you can find some great deals since shops want to clear this all out.
Battery
The biggest hurdle to go off-grid with battery storage in the past has been the battery technology itself. Yes, I know lead-acid batteries have been used, but there are all kinds of issues with it. For example, to maximize cycle life, lead-acid batteries shouldn't be discharged below 50%. This means if you have a 1.2kWh battery (say, a 100Ah, 12V battery) you can only use 600Wh of this battery. In addition, lead-acid is heavy, require maintenance (water level in FLA), like to sit at 100% charge but getting there requires long periods of bulk charging, it has a round trip efficiency in the range of 76% - 85%, substantial Peukert effect, high self-discharge rate (which is why floating exists in charge controllers), overall low cycle life, ventilation needs, etc. If you're in an area with lots of sun, you can somehow make this work, but not over here.
Lithium batteries improve on many of those lead acid negatives, but bring some problems of their own. Typical lithium chemistries such as NMC and NCA rely on, among others, cobalt which besides being a conflict mineral also leads to the 'thermal runaway' property of the battery. This issue means that should a cell be overcharged, the cell can heat up and catch fire, which in turn causes a chain reaction with other cells in the pack. In other words, this chemistry is sometimes described as 'vent with fire'. In the past, a source of batteries were used EV packs (usually from cars involved in an accident). I did consider this as an option, until I found a source of LiFePO4 cells.
Unlike NMC and NCA, LiFePO4 (sometimes LFP) does not use cobalt. This chemistry does not exhibit thermal runaway: even if you overcharge, the cell will bloat and can vent its electrolyte, but it won't spontaneously catch fire or explode while taking the rest of the battery with it in a giant ball of fire. The energy density is lower than that of the other lithium chemistries, but for a stationary application such as this, that is not an issue.
LiFePO4 checks all the boxes that lead-acid doesn't: it can be fully discharged (typically used between 10% and 90%, but you can take it further), long cycle life (>3000 typically), very good round trip efficiency (92% - 98%), very low self-discharge, etc. The one disadvantage is that it can't be charged in freezing temperatures. LTO (Lithium Titanate Oxide) would in theory be better since it can be charged below freezing, but it's much more expensive. Adding a small heating element to the LiFePO4 battery and housing the battery in an insulated box solves the problem as well.
There are several sources of both individual cells (for DIY) and complete LFP batteries available nowadays. Some of the most popular complete battery systems at the time of writing are the $1500 rack mountable EG4 5kWh battery or its pro version. For my needs I would have to get at least 5 of these, meaning a cost of $7500 plus shipping. Getting them in Europe is a bit harder, but there are options. It's not what I went with - in part because these were just not available when I built mine - but also because it can be done cheaper if you go the DIY route and because the rack form factor is not ideal for an insulated design.
The above picture shows one of my batteries, with its semi-completed enclosure missing the lid and insulation. It is made up of 16 x 3.2V 280Ah LiFePO4 cells, making a 48V (actually, 51.2V nominal) battery. It has a BMS (battery management system) which is responsible for monitoring individual cell voltages and can shut off the system in case of problems such as cell voltage too low or too high, over-current, temperature issues, etc. The wiring from the battery is additionally protected with a fast acting Class-T fuse. This is what I paid:
- 32 cells in total, 2 x 16 cell battery packs, for a total of 28kWh: €3215.5
- 2 x JK BMS: €175
- 2 x Class-T fuse and holder: 142€
- The bus bars which I made myself with copper braid: €85 for all 32.
All costs including shipping and taxes, for a total of 3617.5€, or 0.129€ per Wh. Compared to the $1500 off the shelf battery I mentioned, that's less than half the price (that one comes in at 0.27€ per Wh, excluding shipping). Now, keep in mind this was built two years ago. Let's calculate what it would cost today.
Cells can be bought from various vendors, but the ones I mention below have a good reputation, and also have stock in US and/or EU warehouses. Bus bars tend to be included as well, but I'll ignore that. Getting them directly from China can shave off a bit extra, but you have to exercise patience while waiting to get them. Similar to solar panels, shipping costs have increased as well. The vendors:
The latter has a
store front with prices and a warehouse in the States, so let's use their cells for the calculation. I'm using the same 280Ah EVE cells I used, and they cost $147 and have free shipping in the States. This brings the cost of cells to $4704, or €4281 at the time of writing. The
JK BMS sets you back around 109€ each today (the 100A version, same as the one I used on mine), including shipping. The Class-T fuses and bus bars cost the same today as they did when I got them, but the fuses might be a bit difficult to get since there is a shortage at the moment. This brings the total to €4726, or €0.17 per Wh. Still cheaper than the off the shelf one.
The good thing is that, even though prices went up, there are several vendors like the ones I mentioned that ship real, new, grade A, whatever you want to call it, cells. This was not the case when I got mine, and things were more of a shot in the dark back then... However be careful since there are a lot of sellers (especially on Aliexpress) that will sell you crappy cells that don't meet their specification. If you go the DIY route, you should do your research and especially join the
DIY Solar Forum to see if others bought from the vendor and what the results were.
I did leave out the enclosure in my cost calculations. The reason for this is that this is going to be different for everyone. In my case, I need an insulated box with a heating element. Someone else might just be able to put them on a shelf and call it a day. This is part of the flexibility that DIY brings to the table, but I understand it's not for everyone.
Charge controller
The charge controller is responsible for getting the solar panel energy to the battery. It needs to do two important things: 1) convert the solar panel voltage to battery voltage, and 2) handle the amount of power the solar array can produce. In theory, you could find a 10kW charge controller for all the panels. For reasons of redundancy, I split the array into two separate 5kW arrays and have a charge controller for each array. This means that, to handle 5kW, I'm looking at two 100A charge controllers for the 51.2V nominal battery voltage.
The solar panels have a maximum power point voltage (Vmp) of 48.40V and an open circuit voltage (Voc) of 58.08V. These numbers are very important since a charge controller can not just handle any arbitrary voltage. You want to make sure that it can handle the Voc of the panel - but there is another caveat: the voltages are specified at 25°C (77F), but they go up when the temperature drops. I made
a calculator a few years ago for this, and for my specific panels at say, -25°C, the Voc voltage goes up to 65.65V! This means that if you had configured your array with two panels in series and a charge controller that could handle 120V based on the Voc of the datasheet, you would have potentially destroyed the controller once you hit a cold, sunny winter's day. Keep in mind that the voltage of the panel also isn't really dependent on the amount of sunlight: small amounts of light on a cloudy day easily make the full rated Voc voltage on the panels, they just can't deliver their rated power in those conditions.
In principle, you want to configure a solar array to be as high a voltage as you can. The reason for this is losses: you can minimize losses by going higher voltage, and it lowers cabling costs because you can use thinner wire. This is the same reason why high voltage is used on power transmission lines. The problem is that high voltage charge controllers are only now becoming more common in the affordable range. When I was looking at charge controllers, they were practically not available, or would cost a ton of money. The sweet spot seemed to be at 150V (and still seems to be today). This means I could put two panels in series and be within the safe Voc range, and then put these series strings in parallel with each other. Finding a 100A charge controller with these requirements that didn't cost an arm and a leg was still time consuming.
Note: I'm only considering MPPT (maximum power point tracking) charge controllers. PWM (pulse width modulation) is an older technology that only works well when the solar voltage is close to the battery voltage. They tend to handle less power than MPPT controller as well, and typically are less efficient.
In the end I came across a company that has been around for a while, but is not well known in the West: MUST Energy. Their
PC1800F series ticks all the boxes. They're also pretty inexpensive: I paid €485.52 for both of them, including shipping and taxes. I only wish they came in a higher voltage version. Nothing much else I can say about these: power comes in, power goes to the battery, day after day, never an issue. MUST Energy has been around for over 20 years, and dealing directly with them has other benefits I'll get to below.
Inverter
There are two main categories of inverters on the market: low frequency and high frequency inverters. I'm not going into the details of both, it's not important. What is important is to know the main difference between the two: the loads they can power. A high frequency inverter is good for many things, except inductive loads such as motors, compressors, etc. This has to do with the surge current these loads require, and a high frequency inverter generally can not handle large surge loads. A low frequency inverter on the other hand is great at powering these loads, but they tend to have slightly less efficiency and higher no-load idle power consumption. High frequency inverters are generally more complex on the electronics side but do away with the giant, heavy copper transformer that a low frequency inverter uses.
This is all very much simplified, and not 100% correct in all cases, but again, it doesn't really matter. I have tried a high frequency inverter in the past, and let's just say they're fine for certain things but not to run a house off of. I have a ventilation system to power, a well pump, refrigerator, and in summer wood processing tools such as saws, compressors, food dryers, and a wood splitter.... all this tends to be a bit too much for a high frequency inverter.
Since I already found the charge controller from MUST Energy, I looked at their inverters as well. Their low frequency inverters are available in a range from 1kW (Plus series) to 12kW (Pro series). They also include a charger, so batteries can be charged from e.g. a generator. I decided to ask for a quote including shipping for
the 6kW Plus version. Total cost including shipping and taxes came to 683€ - that's very low, considering this thing weighs around 45kg due to a massive toroidal transformer and was shipped by courier service.
Now, did I expect these charge controllers and inverter to work as advertised? At this price point? I seriously had my doubts... but here we are, two years in and no problems at all. Since we are off-grid and when things break you don't want to be out of power for very long, I also inquired about spare parts for the inverter, and ordered a replacement power and control board. They shipped both to me by courier for a total of $150 or so, and they arrived a week later. I practiced replacing both boards in the inverter and that took me about an hour. This whole experience really was beyond my expectations. Incidentally, his is also why I didn't want an all-in-one (inverter + charge controller in one unit): they're much harder to fix and offer no redundancy - when one component fails, everything goes down.
Now, remember what I said about idle power consumption, and this thing isn't an exception: around 50W. This is not an issue in summer, but in winter this adds up. However in winter I don't need this amount of power - no power tools are used, no food dryer is running, and if you remember the power audit, there aren't that many devices to run 24/7 in the first place, none of which are high power. The well pump and sewer fan are the biggest loads, but even those are not huge. Added to that, the well pump is a
Grundfos SQE which has a soft-starter, so there aren't any huge startup power requirements. So, the solution I implemented was running two inverters: the big 6kW one for summer or whenever power is plenty and heavy loads are used, and a small, efficient inverter for winter time. I settled on a Victron device for this: a
Phoenix 48/1200 (the older model, non VE.Direct) that I picked up, new old stock, for 150€. This inverter only has a 6W idle power draw.
Conclusion
Let's tally this all together:
- Solar panels: 4141€
- Ground mounts: 3x €320
- Battery: 3617.5€
- Charge controller: €485.52
- Inverters: 683€ + 150€
- Inverter spare parts: €130
Total: €9846.52. That's a complete DIY system that runs an entire house for a family of four for under 10k€! The cost to have a grid connection to this place would have been more than that. Yes, I should add the generator to this as well for my location, but others that don't have my climate should be able to do the same, or even cheaper if you can manage with half the panels and battery. Cabling cost I also left off, but this does not have a huge impact, and depends a lot on local availability. As for the generator I have: it's a diesel that can run on recycled cooking oil/vegetable oil that I can get almost for free here - so fuel costs are negligible. Since this entry is already very long, I won't go into more details on that right now. Maybe there is room for this (and some other aspects I still have to implement) in a later entry.
Finally, I would like to thank
Will Prowse and Andy from
Off-Grid Garage for sharing this blog on their respective YouTube channels!
Comments
Post a Comment